split-dos site projection issue

[cjh085]

Hello there,

I have been running DOS calculations for magnetic systems, (with LORBIT=11), and using the split_dos code to perform site projections.
If I print out the total DOS from the DOS0 file, I find that it does not equal that which I get from the sum of the DOS data from the separate atoms.

As an example, for NO (ISPIN=2), I get the full 11 electrons in the filled region from DOS0, but the sum of s,px,py,pz (and also the empty d states) for N plus those on O comes to about 9 electrons. This is spin polarized, so the sum is s(up)-s(down)+py(up)-py(down)...etc, due to the sign inversion that split_dos imparts on the spin down states.

Weirdly, the sum from N+O is also missing states at high (antibonding) energies that are present in the DOS0 file.

My input has:
LORBIT=11
NEDOS=1000
ISPIN=2
MAGMOM = 0 1

I attach a plot of DOS0 and N+O-DOS to show this.

Any ideas as to what I could be doing wrong would be great.

Thanks,
Chris

[graeme]

Perhaps this just related to the fact that the projection in done inside a sphere around each atom. The spheres will be characteristic of the atoms but they do not fill space. Depending upon the sphere sizes (RWIGGS) you can have more or fewer electrons in the sum of the projections as compared to the total.